gsd-pi 2.33.0-dev.bafba33 → 2.33.1-dev.235d83a

package/dist/resources/extensions/gsd/tests/context-budget.test.ts

@@ -1,352 +0,0 @@
-/**
- * Unit tests for context-budget.ts — the budget engine.
- * Tests pure functions with dependency-injected fakes.
- * No I/O, no extension context, no global state.
- */
-
-import { describe, it } from "node:test";
-import assert from "node:assert/strict";
-
-import {
-  type BudgetAllocation,
-  type MinimalModel,
-  type MinimalModelRegistry,
-  type MinimalPreferences,
-  type TruncationResult,
-  computeBudgets,
-  truncateAtSectionBoundary,
-  resolveExecutorContextWindow,
-} from "../context-budget.js";
-
-import type { TokenProvider } from "../token-counter.js";
-
-// ─── Test helpers ─────────────────────────────────────────────────────────────
-
-function makeRegistry(models: MinimalModel[]): MinimalModelRegistry {
-  return { getAll: () => models };
-}
-
-function makeModel(id: string, provider: string, contextWindow: number): MinimalModel {
-  return { id, provider, contextWindow };
-}
-
-// ─── computeBudgets ──────────────────────────────────────────────────────────
-
-describe("context-budget: computeBudgets", () => {
-  it("returns proportional allocations for 128K context window", () => {
-    const b = computeBudgets(128_000);
-    // 128K tokens × 4 chars/token = 512K chars total
-    assert.equal(b.summaryBudgetChars, Math.floor(512_000 * 0.15));
-    assert.equal(b.inlineContextBudgetChars, Math.floor(512_000 * 0.40));
-    assert.equal(b.verificationBudgetChars, Math.floor(512_000 * 0.10));
-    assert.equal(b.continueThresholdPercent, 70);
-    assert.equal(b.taskCountRange.min, 2);
-    assert.equal(b.taskCountRange.max, 5);
-  });
-
-  it("returns proportional allocations for 200K context window", () => {
-    const b = computeBudgets(200_000);
-    // 200K tokens × 4 = 800K chars
-    assert.equal(b.summaryBudgetChars, Math.floor(800_000 * 0.15));
-    assert.equal(b.inlineContextBudgetChars, Math.floor(800_000 * 0.40));
-    assert.equal(b.verificationBudgetChars, Math.floor(800_000 * 0.10));
-    assert.equal(b.taskCountRange.min, 2);
-    assert.equal(b.taskCountRange.max, 6);
-  });
-
-  it("returns proportional allocations for 1M context window", () => {
-    const b = computeBudgets(1_000_000);
-    // 1M tokens × 4 = 4M chars
-    assert.equal(b.summaryBudgetChars, Math.floor(4_000_000 * 0.15));
-    assert.equal(b.inlineContextBudgetChars, Math.floor(4_000_000 * 0.40));
-    assert.equal(b.verificationBudgetChars, Math.floor(4_000_000 * 0.10));
-    assert.equal(b.taskCountRange.min, 2);
-    assert.equal(b.taskCountRange.max, 8);
-  });
-
-  it("scales proportionally — 1M > 200K > 128K for all budget fields", () => {
-    const b128 = computeBudgets(128_000);
-    const b200 = computeBudgets(200_000);
-    const b1M = computeBudgets(1_000_000);
-
-    assert.ok(b1M.summaryBudgetChars > b200.summaryBudgetChars);
-    assert.ok(b200.summaryBudgetChars > b128.summaryBudgetChars);
-
-    assert.ok(b1M.inlineContextBudgetChars > b200.inlineContextBudgetChars);
-    assert.ok(b200.inlineContextBudgetChars > b128.inlineContextBudgetChars);
-
-    assert.ok(b1M.verificationBudgetChars > b200.verificationBudgetChars);
-    assert.ok(b200.verificationBudgetChars > b128.verificationBudgetChars);
-
-    assert.ok(b1M.taskCountRange.max >= b200.taskCountRange.max);
-    assert.ok(b200.taskCountRange.max >= b128.taskCountRange.max);
-  });
-
-  it("enforces task count floor (min ≥ 2) at all sizes", () => {
-    for (const size of [128_000, 200_000, 1_000_000, 50_000]) {
-      const b = computeBudgets(size);
-      assert.ok(b.taskCountRange.min >= 2, `min should be ≥ 2 at ${size}, got ${b.taskCountRange.min}`);
-    }
-  });
-
-  it("task count ceiling exists and is bounded", () => {
-    const b = computeBudgets(10_000_000); // very large window
-    assert.ok(b.taskCountRange.max <= 8, `max should be capped, got ${b.taskCountRange.max}`);
-    assert.ok(b.taskCountRange.max >= b.taskCountRange.min);
-  });
-
-  it("handles zero input gracefully — defaults to 200K", () => {
-    const b = computeBudgets(0);
-    const b200 = computeBudgets(200_000);
-    assert.deepStrictEqual(b, b200);
-  });
-
-  it("handles negative input gracefully — defaults to 200K", () => {
-    const b = computeBudgets(-100);
-    const b200 = computeBudgets(200_000);
-    assert.deepStrictEqual(b, b200);
-  });
-});
-
-// ─── truncateAtSectionBoundary ───────────────────────────────────────────────
-
-describe("context-budget: truncateAtSectionBoundary", () => {
-  it("returns content unchanged when under budget", () => {
-    const content = "### Section 1\nSome text.\n\n### Section 2\nMore text.";
-    const result = truncateAtSectionBoundary(content, 10_000);
-    assert.equal(result.content, content);
-    assert.equal(result.droppedSections, 0);
-  });
-
-  it("returns empty string unchanged", () => {
-    const result = truncateAtSectionBoundary("", 100);
-    assert.equal(result.content, "");
-    assert.equal(result.droppedSections, 0);
-  });
-
-  it("truncates at section boundary with ### markers", () => {
-    const content = [
-      "### Section A\nContent A is here.\n",
-      "### Section B\nContent B is here.\n",
-      "### Section C\nContent C is here.\n",
-    ].join("");
-
-    // Budget enough for section A only
-    const sectionALen = "### Section A\nContent A is here.\n".length;
-    const result = truncateAtSectionBoundary(content, sectionALen + 5);
-
-    assert.ok(result.content.includes("### Section A"), "should keep section A");
-    assert.ok(result.content.includes("Content A"), "should keep section A content");
-    assert.ok(!result.content.includes("### Section C"), "should drop section C");
-    assert.ok(result.content.includes("[...truncated"), "should include truncation indicator");
-    // Verify truncation count
-    assert.ok(result.content.includes("truncated 2 sections"), `should show 2 truncated, got: ${result.content}`);
-    assert.equal(result.droppedSections, 2);
-  });
-
-  it("truncates at --- divider boundaries", () => {
-    const content = "Intro text.\n\n---\n\nMiddle section.\n\n---\n\nFinal section.";
-    // Budget enough for intro only
-    const result = truncateAtSectionBoundary(content, 20);
-
-    assert.ok(result.content.includes("Intro text"), "should keep intro");
-    assert.ok(result.content.includes("[...truncated"), "should include truncation indicator");
-    assert.ok(result.droppedSections > 0, "should report dropped sections");
-  });
-
-  it("handles content with no section markers — keeps as much as fits", () => {
-    const content = "A".repeat(200);
-    const result = truncateAtSectionBoundary(content, 50);
-
-    assert.ok(result.content.length < 200, "should be shorter than original");
-    assert.ok(result.content.includes("[...truncated 1 sections]"), "should indicate truncation");
-    assert.ok(result.content.startsWith("AAAA"), "should keep content from the start");
-    assert.equal(result.droppedSections, 1);
-  });
-
-  it("handles content at exact boundary — returns unchanged", () => {
-    const content = "### Section 1\nText here.";
-    const result = truncateAtSectionBoundary(content, content.length);
-    assert.equal(result.content, content);
-    assert.equal(result.droppedSections, 0);
-  });
-
-  it("always keeps at least the first section even if it exceeds budget", () => {
-    const content = "### Long Section\n" + "X".repeat(500) + "\n\n### Short\nY";
-    const result = truncateAtSectionBoundary(content, 10);
-
-    // First section should be present even though it exceeds budget
-    assert.ok(result.content.includes("### Long Section"), "should keep first section");
-    assert.ok(result.content.includes("[...truncated 1 sections]"), "should indicate remaining sections dropped");
-    assert.equal(result.droppedSections, 1);
-  });
-});
-
-// ─── resolveExecutorContextWindow ────────────────────────────────────────────
-
-describe("context-budget: resolveExecutorContextWindow", () => {
-  it("returns configured executor model's contextWindow when found", () => {
-    const registry = makeRegistry([
-      makeModel("claude-opus-4-6", "anthropic", 200_000),
-      makeModel("claude-sonnet-4-20250514", "anthropic", 200_000),
-      makeModel("gpt-4o", "openai", 128_000),
-    ]);
-    const prefs: MinimalPreferences = {
-      models: { execution: "gpt-4o" },
-    };
-
-    const result = resolveExecutorContextWindow(registry, prefs);
-    assert.equal(result, 128_000);
-  });
-
-  it("supports provider/model format in preferences", () => {
-    const registry = makeRegistry([
-      makeModel("gpt-4o", "openai", 128_000),
-      makeModel("gpt-4o", "azure", 64_000),
-    ]);
-    const prefs: MinimalPreferences = {
-      models: { execution: "azure/gpt-4o" },
-    };
-
-    const result = resolveExecutorContextWindow(registry, prefs);
-    assert.equal(result, 64_000);
-  });
-
-  it("supports object format preferences with model + fallbacks", () => {
-    const registry = makeRegistry([
-      makeModel("claude-opus-4-6", "anthropic", 200_000),
-    ]);
-    const prefs: MinimalPreferences = {
-      models: { execution: { model: "claude-opus-4-6", fallbacks: ["gpt-4o"] } },
-    };
-
-    const result = resolveExecutorContextWindow(registry, prefs);
-    assert.equal(result, 200_000);
-  });
-
-  it("falls back to sessionContextWindow when executor model not found", () => {
-    const registry = makeRegistry([
-      makeModel("claude-opus-4-6", "anthropic", 200_000),
-    ]);
-    const prefs: MinimalPreferences = {
-      models: { execution: "nonexistent-model" },
-    };
-
-    const result = resolveExecutorContextWindow(registry, prefs, 300_000);
-    assert.equal(result, 300_000);
-  });
-
-  it("falls back to sessionContextWindow when no execution preference set", () => {
-    const registry = makeRegistry([
-      makeModel("claude-opus-4-6", "anthropic", 200_000),
-    ]);
-    const prefs: MinimalPreferences = { models: {} };
-
-    const result = resolveExecutorContextWindow(registry, prefs, 128_000);
-    assert.equal(result, 128_000);
-  });
-
-  it("falls back to 200K when no session and no executor model", () => {
-    const registry = makeRegistry([]);
-    const prefs: MinimalPreferences = { models: { execution: "missing" } };
-
-    const result = resolveExecutorContextWindow(registry, prefs);
-    assert.equal(result, 200_000);
-  });
-
-  it("falls back to 200K with undefined preferences", () => {
-    const result = resolveExecutorContextWindow(undefined, undefined);
-    assert.equal(result, 200_000);
-  });
-
-  it("falls back to 200K with undefined registry", () => {
-    const prefs: MinimalPreferences = { models: { execution: "claude-opus-4-6" } };
-    const result = resolveExecutorContextWindow(undefined, prefs);
-    assert.equal(result, 200_000);
-  });
-
-  it("ignores models with contextWindow ≤ 0", () => {
-    const registry = makeRegistry([
-      makeModel("broken-model", "test", 0),
-    ]);
-    const prefs: MinimalPreferences = { models: { execution: "broken-model" } };
-
-    const result = resolveExecutorContextWindow(registry, prefs, 128_000);
-    assert.equal(result, 128_000); // falls through to session
-  });
-
-  it("ignores sessionContextWindow ≤ 0", () => {
-    const registry = makeRegistry([]);
-    const prefs: MinimalPreferences = {};
-
-    const result = resolveExecutorContextWindow(registry, prefs, -1);
-    assert.equal(result, 200_000); // falls through to default
-  });
-});
-
-// ─── computeBudgets with provider ─────────────────────────────────────────────
-
-describe("context-budget: computeBudgets with provider", () => {
-  it("anthropic budgets differ from default budgets for same window", () => {
-    const defaultBudgets = computeBudgets(200_000);
-    const anthropicBudgets = computeBudgets(200_000, "anthropic");
-
-    // anthropic uses 3.5 chars/token vs default 4.0
-    // so anthropic totalChars = 200K * 3.5 = 700K vs default 200K * 4 = 800K
-    assert.ok(
-      anthropicBudgets.summaryBudgetChars < defaultBudgets.summaryBudgetChars,
-      `anthropic summary (${anthropicBudgets.summaryBudgetChars}) should be less than default (${defaultBudgets.summaryBudgetChars})`,
-    );
-    assert.ok(
-      anthropicBudgets.inlineContextBudgetChars < defaultBudgets.inlineContextBudgetChars,
-      `anthropic inline (${anthropicBudgets.inlineContextBudgetChars}) should be less than default (${defaultBudgets.inlineContextBudgetChars})`,
-    );
-  });
-
-  it("openai provider matches default budgets (both use 4.0 chars/token)", () => {
-    const defaultBudgets = computeBudgets(128_000);
-    const openaiBudgets = computeBudgets(128_000, "openai");
-
-    assert.deepStrictEqual(openaiBudgets, defaultBudgets);
-  });
-
-  it("anthropic budgets are proportional to 3.5 chars/token", () => {
-    const b = computeBudgets(200_000, "anthropic");
-    // 200K tokens * 3.5 chars/token = 700K chars total
-    assert.equal(b.summaryBudgetChars, Math.floor(700_000 * 0.15));
-    assert.equal(b.inlineContextBudgetChars, Math.floor(700_000 * 0.40));
-    assert.equal(b.verificationBudgetChars, Math.floor(700_000 * 0.10));
-  });
-
-  it("bedrock budgets match anthropic (both use 3.5 chars/token)", () => {
-    const anthropicBudgets = computeBudgets(200_000, "anthropic");
-    const bedrockBudgets = computeBudgets(200_000, "bedrock");
-
-    assert.deepStrictEqual(bedrockBudgets, anthropicBudgets);
-  });
-
-  it("default behavior unchanged when no provider is passed", () => {
-    const b = computeBudgets(128_000);
-    // 128K * 4 = 512K
-    assert.equal(b.summaryBudgetChars, Math.floor(512_000 * 0.15));
-    assert.equal(b.inlineContextBudgetChars, Math.floor(512_000 * 0.40));
-    assert.equal(b.verificationBudgetChars, Math.floor(512_000 * 0.10));
-    assert.equal(b.continueThresholdPercent, 70);
-    assert.equal(b.taskCountRange.min, 2);
-    assert.equal(b.taskCountRange.max, 5);
-  });
-
-  it("task count range is unaffected by provider", () => {
-    const defaultBudgets = computeBudgets(200_000);
-    const anthropicBudgets = computeBudgets(200_000, "anthropic");
-
-    assert.deepStrictEqual(anthropicBudgets.taskCountRange, defaultBudgets.taskCountRange);
-    assert.equal(anthropicBudgets.continueThresholdPercent, defaultBudgets.continueThresholdPercent);
-  });
-
-  it("handles zero input with provider — defaults to 200K", () => {
-    const b = computeBudgets(0, "anthropic");
-    const b200 = computeBudgets(200_000, "anthropic");
-    assert.deepStrictEqual(b, b200);
-  });
-});

package/dist/resources/extensions/gsd/tests/context-compression.test.ts

@@ -1,193 +0,0 @@
-/**
- * Context Compression — unit tests for M004/S02.
- *
- * Verifies that prompt builders respect inlineLevel parameter by
- * inspecting the auto-prompts.ts source for level-aware gating.
- * Cannot call builders directly due to @gsd/pi-coding-agent import
- * resolution — uses source-level structural verification instead.
- */
-
-import test from "node:test";
-import assert from "node:assert/strict";
-import { readFileSync } from "node:fs";
-import { join, dirname } from "node:path";
-import { fileURLToPath } from "node:url";
-
-const __dirname = dirname(fileURLToPath(import.meta.url));
-const promptsSrc = readFileSync(join(__dirname, "..", "auto-prompts.ts"), "utf-8");
-
-// ═══════════════════════════════════════════════════════════════════════════
-// inlineLevel Parameter Presence
-// ═══════════════════════════════════════════════════════════════════════════
-
-const BUILDERS_WITH_LEVEL = [
-  "buildPlanMilestonePrompt",
-  "buildPlanSlicePrompt",
-  "buildExecuteTaskPrompt",
-  "buildCompleteSlicePrompt",
-  "buildCompleteMilestonePrompt",
-  "buildReassessRoadmapPrompt",
-];
-
-for (const builder of BUILDERS_WITH_LEVEL) {
-  test(`compression: ${builder} accepts inlineLevel parameter`, () => {
-    // Find the function signature
-    const sigRegex = new RegExp(`export async function ${builder}\\([^)]*level\\?: InlineLevel`);
-    assert.ok(
-      sigRegex.test(promptsSrc),
-      `${builder} should have level?: InlineLevel parameter`,
-    );
-  });
-}
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Default Level Resolution
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: builders default to resolveInlineLevel() when no level passed", () => {
-  const defaultPattern = /const inlineLevel = level \?\? resolveInlineLevel\(\)/g;
-  const matches = promptsSrc.match(defaultPattern);
-  assert.ok(matches, "should have resolveInlineLevel() fallback");
-  assert.ok(
-    matches.length >= BUILDERS_WITH_LEVEL.length,
-    `should have ${BUILDERS_WITH_LEVEL.length} fallback instances, found ${matches?.length}`,
-  );
-});
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Minimal Level — Template Reduction
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: buildExecuteTaskPrompt minimal drops decisions template", () => {
-  // In the execute-task builder, minimal should only inline task-summary, not decisions
-  assert.ok(
-    promptsSrc.includes('inlineLevel === "minimal"') &&
-    promptsSrc.includes('inlineTemplate("task-summary"'),
-    "execute-task should conditionally include decisions template based on level",
-  );
-});
-
-test("compression: buildExecuteTaskPrompt minimal truncates prior summaries", () => {
-  assert.ok(
-    promptsSrc.includes('inlineLevel === "minimal" && priorSummaries.length > 1'),
-    "execute-task should limit prior summaries for minimal level",
-  );
-});
-
-test("compression: buildExecuteTaskPrompt passes verificationBudget to loadPrompt (#707)", () => {
-  // The execute-task template declares {{verificationBudget}} — the builder must supply it
-  assert.ok(
-    promptsSrc.includes("verificationBudget"),
-    "buildExecuteTaskPrompt should pass verificationBudget in the loadPrompt vars object",
-  );
-  // Verify it computes the budget from computeBudgets
-  assert.ok(
-    promptsSrc.includes("computeBudgets(contextWindow)"),
-    "buildExecuteTaskPrompt should compute budgets from the executor context window",
-  );
-});
-
-test("compression: buildPlanMilestonePrompt minimal drops project/requirements/decisions files", () => {
-  // The plan-milestone builder should gate root file inlining on inlineLevel
-  assert.ok(
-    promptsSrc.includes('inlineLevel !== "minimal"') &&
-    promptsSrc.includes("inlineProjectFromDb(base)"),
-    "plan-milestone should conditionally include project.md based on level",
-  );
-});
-
-test("compression: buildPlanMilestonePrompt minimal drops extra templates", () => {
-  // Full inlines 5 templates, minimal should inline fewer
-  assert.ok(
-    promptsSrc.includes('if (inlineLevel === "full")') &&
-    promptsSrc.includes('inlineTemplate("secrets-manifest"'),
-    "plan-milestone should only include secrets-manifest template at full level",
-  );
-});
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Complete-Slice Level Gating
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: buildCompleteSlicePrompt minimal drops requirements", () => {
-  // Find the complete-slice section and verify requirements gating
-  const completeSliceIdx = promptsSrc.indexOf("buildCompleteSlicePrompt");
-  const nextBuilder = promptsSrc.indexOf("buildCompleteMilestonePrompt");
-  const completeSliceBlock = promptsSrc.slice(completeSliceIdx, nextBuilder);
-  assert.ok(
-    completeSliceBlock.includes('inlineLevel !== "minimal"'),
-    "complete-slice should gate requirements inlining on level",
-  );
-});
-
-test("compression: buildCompleteSlicePrompt minimal drops UAT template", () => {
-  const completeSliceIdx = promptsSrc.indexOf("buildCompleteSlicePrompt");
-  const nextBuilder = promptsSrc.indexOf("buildCompleteMilestonePrompt");
-  const completeSliceBlock = promptsSrc.slice(completeSliceIdx, nextBuilder);
-  assert.ok(
-    completeSliceBlock.includes('inlineLevel !== "minimal"') &&
-    completeSliceBlock.includes('inlineTemplate("uat"'),
-    "complete-slice should conditionally include UAT template based on level",
-  );
-});
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Complete-Milestone Level Gating
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: buildCompleteMilestonePrompt minimal drops root GSD files", () => {
-  const completeMilestoneIdx = promptsSrc.indexOf("buildCompleteMilestonePrompt");
-  const nextBuilder = promptsSrc.indexOf("buildReplanSlicePrompt");
-  const block = promptsSrc.slice(completeMilestoneIdx, nextBuilder);
-  assert.ok(
-    block.includes('inlineLevel !== "minimal"') &&
-    (block.includes('inlineGsdRootFile(base, "requirements.md"') || block.includes('inlineRequirementsFromDb(base')),
-    "complete-milestone should gate root file inlining on level",
-  );
-});
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Reassess-Roadmap Level Gating
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: buildReassessRoadmapPrompt minimal drops project/requirements/decisions", () => {
-  const reassessIdx = promptsSrc.indexOf("buildReassessRoadmapPrompt");
-  const block = promptsSrc.slice(reassessIdx, reassessIdx + 1500);
-  assert.ok(
-    block.includes('inlineLevel !== "minimal"'),
-    "reassess-roadmap should gate file inlining on level",
-  );
-});
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Full Level — No Regression
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: full level preserves all templates and files (no regression)", () => {
-  // Verify the key template names are still present in the source
-  const expectedTemplates = [
-    "roadmap", "decisions", "plan", "task-plan", "secrets-manifest",
-    "task-summary", "slice-summary", "uat", "milestone-summary",
-  ];
-  for (const tpl of expectedTemplates) {
-    assert.ok(
-      promptsSrc.includes(`inlineTemplate("${tpl}"`),
-      `template "${tpl}" should still be present in auto-prompts.ts`,
-    );
-  }
-});
-
-// ═══════════════════════════════════════════════════════════════════════════
-// Import Verification
-// ═══════════════════════════════════════════════════════════════════════════
-
-test("compression: auto-prompts.ts imports resolveInlineLevel and InlineLevel", () => {
-  assert.ok(
-    promptsSrc.includes("resolveInlineLevel"),
-    "should import resolveInlineLevel from preferences",
-  );
-  assert.ok(
-    promptsSrc.includes("InlineLevel"),
-    "should import InlineLevel type from types",
-  );
-});